Evaluating the Efficacy of Vectocardiographic and ECG Parameters for Efficient Tertiary Cardiology Care Allocation Using Decision Tree Analysis

TL;DR

The paper tests whether vectocardiography-based GEH markers, extracted from standard ECGs using the Kors transformation, can improve prediction of cardiovascular events in patients referred to tertiary care. Using four feature sets (S, R, G, SRG) and multiple XGBoost decision trees, the study selects models by AUCPR and AUC, emphasizing high sensitivity for triage. Results show significant GEH differences between those with and without events, with SVG-related features driving the strongest predictive gains; the SRG model achieves the best performance (F$_2$ ≈ 0.62, AUC ≈ 67.6%), while maintaining high sensitivity (~94.12%). The findings support incorporating VCG-derived GEH features into risk stratification to optimize resource allocation, though they call for larger, prospective validation to confirm clinical utility. The approach also highlights the transparency of decision-tree models as compatible with clinical decision-making.

Abstract

Use real word data to evaluate the performance of the electrocardiographic markers of GEH as features in a machine learning model with Standard ECG features and Risk Factors in Predicting Outcome of patients in a population referred to a tertiary cardiology hospital. Patients forwarded to specific evaluation in a cardiology specialized hospital performed an ECG and a risk factor anamnesis. A series of follow up attendances occurred in periods of 6 months, 12 months and 15 months to check for cardiovascular related events (mortality or new nonfatal cardiovascular events (Stroke, MI, PCI, CS), as identified during 1-year phone follow-ups. The first attendance ECG was measured by a specialist and processed in order to obtain the global electric heterogeneity (GEH) using the Kors Matriz. The ECG measurements, GEH parameters and risk factors were combined for training multiple instances of XGBoost decision trees models. Each instance were optmized for the AUCPR and the instance with higher AUC is chosen as representative to the model. The importance of each parameter for the winner tree model was compared to better understand the improvement from using GEH parameters. The GEH parameters turned out to have statistical significance for this population specially the QRST angle and the SVG. The combined model with the tree parameters class had the best performance. The findings suggest that using VCG features can facilitate more accurate identification of patients who require tertiary care, thereby optimizing resource allocation and improving patient outcomes. Moreover, the decision tree model's transparency and ability to pinpoint critical features make it a valuable tool for clinical decision-making and align well with existing clinical practices.

Figures (4)

• Figure 1: Process for obtaining the GEH parameters. 12-lead ECG (A), measurement of the wave intervals and average beat computation. Kors matrix is applied to generate the Frank leads and Vectocardiogram (B) Computation of the vectocardiographic features and Spatial Vector Gradient (C)
• Figure 2: Representation of SVG Elevation and Azimuth angles.
• Figure 3: Overall process to obtain the model representative. (A) The parameter set that will be used for training (B) 50 different instances of XGBoost tree were trained to mitigate the randomness influence. Each instance using the AUCPR as metric (C) The instance with higher AUC is chosen as representative to the model.
• Figure 4: Comparison of the ROC of the models and importance of the features for the winner model